Potential massive error-cloud of *Mecidea prolixa*

General
1

I was about to flag this taxon, but decided that maybe the forum was a better place to start.

Mecidea prolixa is native to Africa, apparently not even widespread in Africa but endemic to the few southernmost countries, yet more than half of iNat’s observations are from North America. I can find no documentation, in print or online, that the species has become established on this continent. I found one article from 1952 mentioning that “On two occasions living specimens of Mecidea prolixa have been intercepted by quarantine inspectors at United States ports, each time in shipments of grass seed from South Africa.” Absolutely no mention of the species anywhere in North America before that or since.

Usually iNat flags non-native species with an exclamation point. M. prolixa is African, but the North American records have no flags indicating its non-native status.

The species of Mecidea are similar enough to each other to make ID from photos difficult and confusion easy.

Of iNat’s 269 observations of the species, 171 are from North America; 24 of these are “research grade”. I’ve reviewed all of those “research grade” observations, and not a single one of them gives any justification for an identification of prolixa over the three species known from the continent (M. major, M. minor, and the Caribbean M. longula which apparently has been introduced to Florida). None of reporters or identifiers even display any awareness that there are any other species in the genus, much less explaining why they picked the African-origin one over the known American species.

The situation reeks of a lot of people picking a species name from the auto-ID suggestions list without looking at the range map or reading information about the species, other people clicking “agree”, and all of it snowballing into iNat’s data for this species being massively corrupted with errors.

I’ll tag some people who specialize more than I do in the species, genus, or family involved, to see if any of them know something I don’t about this M. prolixa situation… @ameeds , @wongun , @roland_lupoli , @kgrebennikov , @seedballtheearth , @dhjacobs , @rjpretor , @traianbertau Can any of y’all explain this situation? Did M. prolixa actually establish a North American population without anyone talking about it, or is it necessary to spam the 171 records in question with ID corrections?

2

no need to flag the taxon. if you think it’s wrong, then just add some correcting IDs.

somebody has to actually add establishment means records, and anybody can add these, but no one has added any establishment means records yet for the taxon for any place. however, even adding establishment means of non-native wouldn’t say that the taxon couldn’t be present in a particular place, only that it would be non-native if found there.

that’s likely true, but this wouldn’t be the only situation where this has occurred, and really the best solution is to go and just correct the errors. once corrected, then that may fix the geomodel used by the computer vision the next time it’s updated.

3

Geomodel for that sp

image
image730×403 44.6 KB

Leave a brief copypasta as you correct the wrong IDs, and that ripples out to identifiers and observers in USA.

4

I never found this type of map. Is there a way to open those through the menu?

5

there’s a link on the taxon page:

image
image649×874 138 KB

6

Go to the taxon page
Click About
Then

image
image532×295 20.9 KB

and click that blue link – which it took me a while to realise – is the Geomodel for this taxon ! It sounds like a generic link to – what even IS the Geomodel ?

(Sorry – duplicated)

7

In similar situations I typically do what y’all have recommended, and that’s probably what I’ll ultimately wind up doing here, correcting each observation with a quick copypasta. But the “similar situations” previously all involved a species where 98% or so of iNat’s observations were on the continent where the species was known to occur, and I was just dealing with the 2% on other continents. I’ve never before encountered a situation where the IDs on the wrong continent outnumbered those on the right one, outside of well-known, thoroughly documented invasions. Hence my tagging the list of hardcore hemiptera-heads above, just in case something new has happened that nobody’s managed to publish or publicize yet. If a new invasion has really happened - and the 1952 article illustrates that the potential is certainly there - then I don’t want to inadvertently conceal it. If none of the bug-perts give me reason to believe those IDs, I’ll do as y’all recommend in the near future, maybe as soon as later tonight…

8

looking at the identifications in the link i noted in my previous post, it looks like most of the IDs were done with the assistance of computer vision or were agreements to such IDs by random users. it’s not to say that none of them know what they’re doing, but it seems like usually when there’s a new detection in an unusual place, people are quick to get excited and add comments, which doesn’t seem to be the case here. there are a handful of folks who have made multiple identifications of the taxon, and you could probably reach out to those folks to see what their process was. it’s possible that by doing so, they’ll withdraw their identifications, which will fix a not insignificant portion of the observations.

9

Maybe try PM to your bug-perts, before you push back the ‘wrong IDs’.
Those on your list are mostly not active on the Forum.
roland lupoli is mostly Europe, he and dhjacobs (the most active identifier) are greyed out.

(I did look at the list of identifiers for USA, just a few IDs each
For South Africa the top 3 are active taxon specialists)

10

@opihiman – Another thing to fix while you’re working on this: The Wikipedia article for Mecidea lists 8 species, not including Mecidea prolixa.

11

Certainly another case of CV error. Stink bugs of North America were catalogued by Rider and Swanson in 2021 (https://doi.org/10.11646/zootaxa.5015.1.1) and three species were known: longula, major and minor. Mecidea prolixa still appears to be only known from southern Africa. There are 16 Mecidea species worldwide per the Rider and Swanson catalogue.

In any case, the genus is difficult to ID to species from photos.

I’ve tried to push back on similar CV errors, but the CV keeps suggesting extralimital species. Unfortunately, it seems like a losing battle unless someone has the time to constantly watch those taxa.

12

I encountered a similar situation with Scolypopa australis, which is well known as an invasive in New Zealand. I discovered it in Hawaii and went to look where else it may have been introduced, and found erroneous observations all over the world. Not the majority, but still a lot. The problem was that once the CV was trained on it, it started identifying every remotely similar fulgoroid nymph as that, since it was the only one in the model.

And yes, out of dozens of incorrect observations I did find one that was a real new introduction, in Buenos Aires. After several years it seems to be still very highly localized there.

13

The big problem is there is a delay between clearing up the observations and the CV learning the species does not occur there. So if you clean it up it will keep suggesting the taxon for a couple months. I think if you are vigilant about doing occasional sweeps for that period the CV will learn the species does not occur there and stop recommending it, so its not a permanent losing battle at least…

14

Would it be impossible to split the two populations and just revert the American ones to genus via taxon swap? Is that impractical?

15

They are more likely to respond on a flag on the species, or on a relevant observation, since many iNat users aren’t active on the forum.

This is not an uncommon situation. I’ve usually used a flag on the species to raise awareness of the CV feedback loop and coordinate a response, particularly if it involves thousands of observations since that takes some effort to clear out (intense effort at first, and then checking back for the next ~6 months until it’s out of both website and app versions of the CV/geomodel).

16

You can correct 171 observations in one batch using the iNaturalist Metadata Tool, if you change the search URL under Identify to show all the observations on one page. Go back and do it again every few weeks until the geomodel catches on.

17

Many thanks for that article Brad! Definitely useful. I think I’m convinced enough to start spamming corrections, though might not really have time until tomorrow…

18

When you come up with a good explanation to “spam,” I’ll borrow it to help with New Mexico

19

Just tackled the handful from Texas. Used this Copy-paste message highlighting the bugguide advanced search with that genus from the state of Texas:

Probably should rest at genus:
https://www.bugguide.net/adv_search/bgsearch.php?user=&taxon=34981&description=&location[]=TX&county=&city_location=&adult=&immature=&male=&female=&representative=

20

Finished with the United States mis-ID’s…

I think one of the challenges in this case is that it’s pretty difficult to get to a species ID with this genus in general, so Mecidea prolixa is the only ‘option’ for a species suggestion, so folks are a bit hasty. I’m sure this exists in a lot of cases with more obscure species.
iNat is still a work in progress! :)